4338
J. Phys. Chem. B 1999, 103, 4338-4341
Self-Association of Rifamycin B: Possible Effects on Molecular Recognition Daniel W. Armstrong,* Jeffrey Schneiderheinze, and Usha Nair Department of Chemistry, UniVersity of MissourisRolla, Rolla, Missouri 65409
Linda J. Magid Department of Chemistry, UniVersity of Tennessee, KnoxVille, Tennessee 37996
Paul D. Butler NIST Center for Neutron Research, National Institute of Standards and Technology, Gaithersburg, Maryland 20899 ReceiVed: January 22, 1999
The macrocyclic antibiotic rifamycin B was found to be highly surface-active and to aggregate in aqueous solution. The aggregational behavior was studied using small-angle neutron scattering (SANS). Aqueous solutions of rifamycin B showed pronounced scattering at both small-length scales (Q g 0.1 Å-1) and at large-length scales (Q e 0.1 Å-1). The larger association colloids appear to be rather open low-density aggregates. The addition of 10% 2-propanol greatly reduces the number and size of the aggregates. Somewhat higher amounts of alcohol appear to completely suppress or eliminate aggregation. The suppression of aggregation coincides with the appearance and enhancement of enantioselective association between rifamycin B and a variety of chiral amino alcohols. It appears that the self-aggregation of rifamycin B may be a factor that controls its ability to differentiate between enantiomers in aqueous and hydro-organic solutions.
Introduction The rifamycin family of antibiotics are a subgroup of the ansamycins that includes streptovaricins, halomycins, mycotrienins, actomycin, and several other structurally related antibiotics.1 The general name ansamycin refers to a specific macrocyclic structure that consists of a chromophoric moiety (generally aromatic) spanned by an aliphatic chain or bridge.1,2 An example of the ansamycin structure is shown in Figure 1 for rifamycin B. Note that the aliphatic bridge must link to nonadjacent positions of the chromophore. This particular molecule has nine stereogenic centers and exists as a dibasic acid with pKa’s of approximately 2.8 and 6.7.3 It has a molecular weight of 755.8 and is slightly soluble in water (more so when ionized) as well as in the smaller alcohols and acetone.3,4 The rifamycins are particularly important antituberculosis compounds5,6 and are promising agents for the treatment of leprosy7 rheumatoid arthritis and other chronic arthritic disorders.8,9 Rifamycins also have been shown to be unusual but effective chiral selectors for high-efficiency separations of enantiomeric compounds (particularly those that contain amine functionalities).4 Rifamycin B was the initial macrocycle in its family to be isolated and characterized. It is produced by fermentation as are other ansamycins (although they can be synthetically modified as well). Originally, it was thought that rifamycin B was produced by Streptomyces mediterranei. Later, on the basis of its cell wall chemistry, the organism was reassigned to the genus Amycolctopsis.1 Previously, in the course of evaluating enantioselective interactions between rifamycin B and a number of aminoalcohol * To whom correspondence should be sent.
drugs, two interesting observations were made (one of which was unique).4 First, aqueous solutions of rifamycins readily produced foam when agitated or stirred. This indicated that rifamycin B may be surface-active. A review of the rifamycin literature revealed only a few corroborating observations of this type. These observations were in the purview of biotechnology/ engineering where it was found that large-scale reactors for the biotransformation of rifamycins sometimes had a foaming problem and that the addition of small amounts of an antifoaming agent was beneficial.10,11 The second observation was more puzzling. No enantioselective interactions between rifamycin B and the chiral aminoalcohols were observed in strictly aqueous or buffered solutions. However, when ∼10% or more of a miscible organic solvent was added to the solution (e.g., methanol, ethanol, 1-propranol, 2-propranol, or acetonitrile), enantioselective binding was observed.4 In many previous capillary electrophoresis (CE) studies (with other chiral selectors) it was found that addition of small amounts of miscible organic solvents could change the degree of enantioselective binding (most often decreasing it), but enantioselectivity was always observed in the optimized aqueous systems.12,13 If, indeed, rifmaycin B is surface-active and concentrates at the air-water interface, it is reasonable to propose that it may self-associate in aqueous solution. Just as micellization of ionic surfactants in water is impeded by increasing concentrations of miscible alcohols, rifamycin B self-assembly could be impeded as well. If so, then it appears that molecular chiral recognition by rifamycin B requires (at least in part) a significant fraction of monomers in solution. In this work, experiments are done and data generated to test the hypothesis outlined above. To our knowledge, there has never been a report on the selfassociation of rifamycin B or possible deleterious effects of self-
10.1021/jp990279+ CCC: $18.00 © 1999 American Chemical Society Published on Web 04/30/1999
Rifamycin B
J. Phys. Chem. B, Vol. 103, No. 21, 1999 4339
Figure 2. Plot of surface tension (distilled/deionized water) versus concentration of rifamycin B.
Figure 1. Structure of rifamycin B (top) and space-filling model (bottom) showing its configuration and 3-D shape. The atoms are colorcoded as follows: carbon (black), oxygen (red), hydrogen (white), and nitrogen (blue).
aggregation on molecular recognition. The effects of rifamycin B concentration and ionic strength also were investigated. Experimental Section Preparation of Rifamycin B Samples for Small-Angle Neutron Scattering (SANS) Experiments. Deuterium oxide (99.9 atom % D), 2-propanol-d8, sodium deuterioxide (40 wt % in D2O), and sodium dihydrogen phosphate were purchased from Aldrich (Milwaukee, WI). Rifamycin B was purchased from Sigma (St. Louis, MO). Samples of 0.05 and 0.1 M phosphate buffers were prepared by dissolving appropriate amounts of NaH2PO4 in D2O and adjusting to a pD of 7.00 using NaOD. Samples of 20, 15, and 10 mM rifamycin B samples were then prepared by dissolving appropriate amounts of rifamycin B in either a phosphate buffer or a phosphate buffer/2-propanol-d8 mixture (v/v). The samples were then transferred to a quartz sample container for SANS analysis. Small-Angle Neutron Scattering. The SANS measurements were performed on the NG-7 SANS spectrometer at the Cold Neutron Research Facility, National Institute of Standards and Technology, Gaithersburg, MD. The range of scattering, Q, used for the majority of the samples was 0.01-0.45 Å-1. Q is defined as (4π/λ)sin θ, where 2θ is the scattering angle and the incident neutron wavelength λ is 5 Å. For 20 mM rifamycin B in 0.05 M phosphate buffer scattering data were collected at lower Q
(Qmin - 0.002 Å-1, λ ) 10 Å). The samples were contained in quartz spectrophotometric cells of 5 mm path length and mounted in a thermostated cell holder held at 25.0 ( 0.1 °C. Each two-dimensional set of raw scattering data was corrected for detector background and sensitivity and for scattering from the empty cell and then radially averaged.15 The resulting I(Q)’s were converted to absolute intensities (in cm-1) using precalibrated secondary standards provided by CNRF, NIST. The final step consisted of the Q-by-Q subtraction of the (flat) background scattering due to the aqueous buffer/alcohol solutions. Experimental radii of gyration (Rg) were obtained from the corrected data over a range of Q from 0.12 to 0.35 Å-1 by making use of the Guinier approximation: ln I(Q) ∝ -Q2Rg2/ 3. Surface Tension Measurements of Rifamycin B Solutions. Rifamycin B solutions were prepared by dissolving appropriate amounts of rifamycin B in 0.1 M NaH2PO4 (pH ) 7.00). The samples were then transferred to a shallow watch glass, and the surface tension was measured using a surface tensiometer model 20 (Fisher Scientific Co.). Between each measurement, the platinum ring of the surface tensiometer was rinsed and flamed in a Mekker burner to ensure the removal of all residual materials. Results and Discussion Rifamycin B does not have a structure that is characteristic of traditional surfactant molecules (i.e., a hydrophilic “head” group and a hydrophobic “tail”).14 Nor is it completely similar to bile salt surfactants, i.e., steroids with a hydrophobic side and a hydrophilic edge.14 However, the space-filling structure of rifamycin B shows a bent, disk-shaped molecule with distinct hydrophobic and hydrophilic regions (Figure 1). Furthermore, as shown in Figure 2, the addition of small amounts of rifamycin B to water lowers the surface tension drastically as would be expected of a more conventional surfactant. Although the simple lowering of surface tension of water and foaming behavior are no guarantee that a molecule will self-associate, these properties certainly warranted further investigation into the possibility. Rifamycin B is anionic at pH 7. Its solutions are intensely colored. Consequently, it is not feasible to use light scattering as an experimental tool for the investigation of its aggregation behavior. Small-angle neutron scattering (SANS) is not affected by the light absorption of a sample. Furthermore, it is sensitive to the length scales that would occur in solutions containing monomeric and oligomeric rifamycin species. SANS Studies. Small-Length Scales. All rifamycin B solutions studied show scattering at Q g 0.1 Å-1. This is
4340 J. Phys. Chem. B, Vol. 103, No. 21, 1999
Armstrong et al.
Figure 3. Small-length neutron scattering data for D2O solutions containing 20 mM rifamycin B in 0.05 M phosphate buffer (pH 7.0). Note the flat profile for solutions containing 30% 2-propanol (4).
TABLE 1: Radii of Gyration of Rifamycin B Aggregatesa [rifa B], mM in 0.05 M phosphate buffer 20
in 0.1 M phosphate buffer 10 15 20
a
wt % 2-propanol
Rg, Å
0 10 20 30 40
5.82 4.92 3.93 3.89 4.08
0 0 0 10 20 30 40
4.80 5.08 5.92 4.57 4.07 4.41 4.57
Error bars on the Rg values are ( 0.3 Å.
characteristic of species having a radius of gyration (Rg) in the 4-6 Å range. Figure 3 shows two typical scattering curves. For 0.05 and 0.1 M phosphate buffer containing 20 mM rifamcyin B, the Rg’s are 5.8 and 5.9 Å, respectively (Table 1); at 10% 2-propanol, the Rg’s are 4.9 and 4.6 Å, respectively. As Table 1 makes clear, in 0.05 M buffer further addition of alcohol (20-40%) causes the Rg to plateau for 20 mM rifamycin B at 4.0 ( 0.1 Å, while in 0.1 M buffer, there is somewhat greater dispersion, with fitted Rg’s in a range from 4.1 to 4.6 Å. For 0.1 M buffer without 2-propanol, the scattering curves for 10 and 15 mM rifamycin B produce Rg’s of 4.8 and 5.1 Å, respectively. Crystal structures and molecular modeling of rifamycin B show an aromatic region with a protruding carboxylatecontaining side chain and an aliphatic “belt” or side chain roughly perpendicular to the aromatic moiety (Figure 1). This molecule is somewhat Janus-faced, having a relatively more hydrophobic side and a hydrophilic one. For a discussion of the small Rg’s reported in Table 1, it is sufficient to ignore the small hydrophobic cavity inside the belt and to approximate rifamycin B monomers as ellipsoids of revolution with three semiaxes, a, b, and c, giving
Rg2 ) (a2 + b2 + c2)/5 From the modeling, we take dimensions for the full axes of 14.3 and 9.5 Å (length and width) and 6.2 Å (the “thickness”,
Figure 4. Plots showing the effect of increasing 2-propanol concentration on the small-length neutron scattering curves for D2O solutions of 20 mM rifamycin B in 0.1 M phosphate buffer (pH 7.0).
a compromise between the side of the molecule with the aromatic moiety plus side chain and the edge of the “belt”). These dimensions result in an Rg for the monomers of 4.1 Å. If the monomers were to dimerize via interaction between the more hydrophobic faces of their “belts”, the 9.5 Å axis would be unaffected, while the thickness would approximately double and the long axis would increase by ca. 5 Å. This results in a dimeric object having an Rg of 5.6 Å. An alternative mode of dimerization, involving the faces of the aromatic moieties (“endto-end” dimerization), would produce an object of dimensions 30 Å × 9 Å × 4.5 Å and an Rg of 6.9 Å, considerably larger than the values observed for rifamycin B in the absence of 2-propanol. The face-to-face belt dimers have the aromatic moieties at opposite ends of the dimers’ longest axis, available for further aggregation in solution; end-to-end dimers have the hydrophobic faces of the belts available for further aggregation. Observations concerning scattering from longer-length scales are made below. Face-to-face dimerization of the belts, as opposed to end-toend dimerization, holds the potential to have more impact on the macrocycle’s conformation and hence on molecular recognition. However, the differences between calculated Rg’s for the two possible dimers and the experimental Rg’s observed in the absence of 2-propanol are large enough to prevent a clear conclusion as to which dimer may predominate. Of course, other small aggregates such as trimers, etc. may contribute at Q g 0.1 Å-1 as well. The increase in Rg with increasing rifamycin B concentration in buffer alone is suggestive of indefinite association, with higher n-mers progressively becoming more important. In addition, the I(Q)’s observed for these three solutions at Q e 0.1 Å-1 increase much more rapidly with increasing concentration than the expected linear scaling that a shared distribution of aggregate sizes would suggest. Large-Length Scales. In the absence of 2-propanol, there is significant scattered intensity at Q e 0.1 Å-1; Figure 4 illustrates this for 20 mM rifamycin B in 0.1 M phosphate buffer. The low-Q scattering is greatly reduced already at 10% alcohol and disappears completely by 30% alcohol (Figure 4) at both buffer concentrations. Attempts to fit these data assuming either cylindrical or disklike morphologies failed. The sample showing the most intense low-Q scattering, 20 mM rifamycin B in 0.05 M phosphate buffer, was investigated
Rifamycin B
J. Phys. Chem. B, Vol. 103, No. 21, 1999 4341 chiral recognition between rifamycin B and a variety of chiral aminoalcohol drugs.4 There are at least two possible scenarios that can be proposed on the basis of this information. One is that the self-aggregation of rifamcyin B may, in some way, impair its ability to differentiate enantiomers of aminoalcohols in aqueous solution. An alternative scenario is that the added alcohol plays some other beneficial (but unknown) role in molecular recognition and that its concurrent inhibition of aggregation is coincidental. Note that the self-aggregation of rifamycin B does not need to completely disrupt the binding of a chiral guest molecule to the rifamycin B.4 Simply altering one of the simultaneous interactions necessary for stereoselective association is sufficient to alter chiral molecular recognition. Hence, one possible interpretation of the results of this study is that they provide some evidence for aggregational mediated chiral molecular recognition. To our knowledge, this is the first report on the aggregational behavior of rifamycin B.
Figure 5. Long-length neutron scattering data for a D2O solution of 20 mM rifamycin B with 0.05 M phosphate (pH 7).
down to 0.002 Å-1. Figure 5 shows that there is a wide range of Q over which I(Q) displays a power-law behavior; the appearance of the scattering curve is reminiscent of data for polymeric fractal materials such as polymer foams or silica aerosols. These materials have multiple length scales, from micrometers to angstrons, and scattering curves having both power-law and Guinier regimes. The Guinier regimes result from Rg’s of the overall structure and from various structural subunits; in each case the power-law decay comes after the exponential regime.17-19 Unfortunately, the Q range for our rifamycin B samples cannot be further extended by either light scattering (intensely colored solutions) or by ultralow SANS or X-ray scattering (insufficient scattered intensity). As a result, we are unable to obtain an Rg for the overall structure but apparently see only its power-law regime. The observed low-Q dependence, roughly I(Q) ∝ Q-2, suggests a mass fractal dimension of 2. This relatively low mass fractal dimension implies that the aggregate has a somewhat open structure of low density. Conclusions Despite having a molecular structure unlike those of other traditional surfactants, rifamycin B both concentrates at the airwater interface, depressing surface tension, and shows selfaggregation in aqueous solutions. Neutron scattering studies have identified both small aggregates, probably concentration-dependent indefinite association type, monomer, dimer, trimer, etc., and much larger aggregates with a relatively open structure of low density. Small amounts of 2-propanol (∼10%) greatly reduces aggregation. At higher concentrations of 2-propanol (∼30-40%) there is little evidence of aggregation. The disappearance of aggregation with increasing concentrations of alcohol seems to correspond to the appearance of molecular
Acknowledgment. Support by the National Institute of Health (NIH R01 GM53825-02) is gratefully acknowledged (D.W.A., J.S., and U.N.). Additional support (L.J. M.) from the NSF, Chemistry Division (CHE-9729433), is also acknowledged. Portions of this material are based on activities supported by the National Science Foundation under Agreement No. DMR-9423101. Identification of certain equipment or materials does not imply recommendation by NIST. References and Notes (1) Chiao, J. S.; Xia, T. J.; Mei, B. G.; Jin, Z. K.; Gu, W. L. Genetics and Biochemistry of Antibiotic Productions in Biotechnology; Vining, L. C., Stuttard, C., Eds.; Butterworth-Heinemann: London, 1995; Vol. 28, Chapter 13, pp 477-491. (2) Prelog, V.; Oppolzer, W. HelV. Chim. Acta 1973, 56, 2279. (3) Sensi, P.; Greco, A. M.; Ballotto, R. Antibiot. Ann. 1959, 60, 266. (4) Armstrong, D. W.; Rundlett, K.; Reid, G. L., III. Anal. Chem. 1994, 66, 1690. (5) Fox, W. Bull. Int. Union Tuberc. 1985, 60, 40. (6) Noordeen, S. K.; Godal, T. Brit. Med. Bull. 1988, 444, 523. (7) Shepard, C. C. Lepr. ReV. 1981, 52, 299. (8) DiGiulo, A.; Barracchini, A.; Cellai, L.; Martuccio, C.; Amicosante, G.; Oratore, A.; Pantaleoni, G. Int. J. Pharm. 1996, 144, 27. (9) Caruso, I. J. Int. Med. Res. 1997, 25, 307. (10) Lee, G. M.; Choi, C. Y.; Park, J. M.; Han, M. H. J. Chem. Technol. Biotechnol. 1984, 35B, 3. (11) Banerrjee, V. C.; Saxeno, B.; Chisti, Y. Biotechnol. AdV. 1992, 10, 577. (12) Wren, S. A. C.; Rowe, R. C. J. Chromatogr. 1992, 603, 235. (13) Ferguson, P. D.; Goodall, D. M.; Van der Wal, S.; Duchateau, A. L. L.; Everaerts, F. M. J. Chromatogr. A 1996, 745, 61. (14) Fendler, J. H.; Fendler, E. J. Catalysis in Micellar and Macromolecular Systems; Academic Press: New York, 1975. (15) NG3 and NG7 30 meter SANS Instruments Data Acquisition Manual, NIST Center for Neutron Research, Sept. 1997. (16) Brufani, M.; Fedeli, W.; Biacomello, G.; Vaciago, A. Experientia 1964, 20, 339. (17) Beaucage, G.; Schaefer, D. W. J. Non-Cryst. Solids 1994, 797, 172. (18) Beaucage, G.; Ulibarri, T. A.; Black, E. P.; Schaefer, D. W. Organic Hybrid Materials; Mark, J. E., Lee, C. Y. C., Bianconi, P. A., Eds.; ACS Symposium Series 585; American Chemical Society: Washington, DC; p 97. (19) Beaucage, G. J. Appl. Crystallogr. 1996, 29, 134.
